ASoC: max98090: request IRQF_ONESHOT interrupt
PATCH 1/1 ASoC: max98090: request IRQF_ONESHOT interrupt
request_threaded_irq() rejects calls which both do not specify a handler (indicating that the primary IRQ handler should be used) and do not set IRQF_ONESHOT because the combination is unsafe with level-triggered interrupts. It is safe in this case, though, since max98090 IRQs are edge-triggered and the interrupts aren't ACK'ed until the codec's IRQ status register is read. Because of this, an IRQF_ONESHOT interrupt doesn't really make a difference, but request one anyway in order to make request_threaded_irq() happy.
--- sound/soc/codecs/max98090.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
sound/soc/codecs/max98090.c +1 -1
@@ -2247,7 +2247,7 @@ static int max98090_probe(struct snd_soc_codec *codec)dev_dbg (codec->dev,"irq = %d\n" , max98090->irq); ret =request_threaded_irq (max98090->irq,NULL , - max98090_interrupt,IRQF_TRIGGER_FALLING , + max98090_interrupt,IRQF_TRIGGER_FALLING |IRQF_ONESHOT ,"max98090_interrupt" , codec);if (ret <0 ) { dev_err(codec->dev, "request_irq failed: %d\n",
Signature · 1 line
1.8.2.1
Mark Brown
maintainerQuoted 4 lines · Andrew Bresticker
request_threaded_irq()rejects calls which both do not specify a handler (indicating that the primary IRQ handler should be used) and do not setIRQF_ONESHOTbecause the combination is unsafe with level-triggered interrupts. It is safe in this case, though, since max98090 IRQs are
Applied, thanks. If request_threaded_irq() is rejecting these that's not ideal - some threaded interrupts can only be generated from within threaded context and therefore can't have primary handlers. Unless it only rejects in the case where the parent IRQ is a hard IRQ of course.